As information continues to emerge about COVID-19, researchers and companies are trying new approaches to map the outbreak.
But how to accurately display and project the spread of the disease has proven difficult, particularly given the limited number of tests available to confirm where cases of COVID-19 actually are.
Getting detailed data into coronavirus hot spots is something the White House has said will help to inform response to the pandemic. The Trump administration is reportedly planning to issue guidelines for how, in concert with expanded testing efforts, governors can categorize counties as high-, medium- or low-risk, as part of an effort to ease social distancing measures.
Last week, Dr. Deborah Birx, the coronavirus response coordinator for the White House's coronavirus task force, referenced disease mapping as a way to better target interventions.
"If we geographically get specific data by ZIP codes and counties, we'll be able to approach this in a very laser-focused way, making sure that what we're doing in each of those areas is absolutely appropriate for where they are in their own little bell-shaped curve," Birx said at a press briefing last Tuesday.
Researchers across the U.S. have been tackling projects related to mapping and projecting the spread of COVID-19 since the disease's emergence.
Some projects, like an online dashboard developed at Johns Hopkins University, map COVID-19 cases worldwide by compiling data from the World Health Organization, the U.S. Centers for Disease Control and Prevention, and state and national health departments, as well as information from organizations like local news media.
Other efforts, like a COVID-19 hospital impact model from Penn Medicine, help hospitals project how many inpatients, intensive care patients and patients on ventilators they're expected to see in the coming weeks, based on current COVID-19 cases in the region and that a hospital is treating.
Monitoring those types of tools has proven useful for Anne Arundel Medical Center, said Dave Lehr, the system's chief information officer.
"We looked at what (COVID-19 cases) would do to our volumes over the course of the next few months," Lehr said. "When it comes to just our day-to-day planning, I think that figuring out how many people are going to be hospitalized in each hospital is the top priority."
The Annapolis, Md.-based regional health system's analytics department has used available data and projections to help build their own models, which help Anne Arundel's incident command team proactively think about how to manage supply over the coming weeks. It's been particularly helpful when it comes to projecting the need of supplies where hospitals are increasingly facing shortages, like personal protective equipment and ventilators, Lehr said.
But mapping the spread of COVID-19 has its challenges.
There are limited tests available to confirm cases, and scientists are still learning more about how the disease spreads every day. It's unclear how many Americans could be infected with the virus but haven't experienced symptoms yet, since COVID-19's incubation period—the length of time between when someone is infected with the virus and when they start displaying symptoms—could be up to two weeks, according to early estimates. And some could be infected but asymptomatic.
The traditional way of mapping communicable diseases like COVID-19 tends to rely on first determining those who have the disease through testing, and then identifying those at risk, such as partners or coworkers who have come into contact with them, said Dr. Georges Benjamin, executive director of the American Public Health Association.
"That's still important to do," he said. "It's effective."
But in light of limited ways to confirm cases of COVID-19, researchers and companies have turned their attention to other ways to estimate the spread of the disease without testing data. Those emerging surveillance methods could supplement information on testing alone.
In recent weeks, companies have put out tools that purport to map COVID-19 spread through a range of approaches, like monitoring searches for disease information, social media posts and even temperatures from smart thermometers. The White House has reportedly asked tech giants like Google and Facebook about using aggregated location data from smartphones for public health surveillance related to COVID-19.
Benjamin called out Flu Near You, a crowd-sourced project that uses syndromic surveillance—a method in which researchers monitor reports of symptoms of a disease before they're confirmed—as a promising approach to detect outbreaks earlier.
Flu Near You was developed by epidemiologists at Boston Children's Hospital, Harvard University and the Skoll Global Threats Fund in 2011. In the wake of the COVID-19 outbreak, researchers from Boston Children's Hospital and Harvard Medical School built off of the project to create a separate tool that monitors COVID-19, which it launched last week.
COVID Near You—and its influenza counterpart—are hosted on public websites, which anyone can visit and enter information like ZIP code, gender, age and how they're feeling to help researchers map areas with a concentration of COVID-19 symptoms.
Given lack of testing makes it difficult to identify confirmed cases, the project could play an important role in COVID-19 response, said John Brownstein, co-founder on the project and chief innovation officer at Boston Children's Hospital, as well as an epidemiologist.
"We need to understand the hotspots," he said. "We need to understand how interventions are working."
More than 100,000 people have submitted data through the COVID Near You website already.
There's also interest in predicting areas likely to be hit hard by COVID-19. Using data on demographic and socio-economic factors, some companies are working to develop tools that they say identify areas particularly vulnerable to COVID-19 spread, based on possible risk factors and social determinants of health.
That could help hospitals and health agencies prioritize outreach to at-risk populations, such as campaigns reminding people in specific neighborhoods or counties to stay indoors or wash their hands, said Dr. John Showalter, chief product officer at Jvion, an analytics company that's mapped where it expects patients are most likely to experience hospitalizations or mortality if they contract COVID-19.
But it's tough, since typical outreach strategies might rely on sending messages through a patient portal—and many vulnerable areas also lack access to home computers or internet. In a 2018 report, the Federal Communications Commission found that 31% of people in rural America, for example, lacked access to wired broadband that met the agency's speed benchmark.
Other common strategies, like door-to-door outreach, aren't feasible in light of the COVID-19 crisis.
That's something hospitals need to keep in mind when targeting preventive campaigns toward high-risk areas, according to Benjamin. He suggested phone calls, text messages and radio could prove useful strategies.
"You certainly have to communicate a little differently," Benjamin said.
